{
  "id": 3350,
  "url": "https://arxiv.org/abs/2606.01139v3",
  "title": "SkillRevise: Improving LLM-Authored Agent Skills via Trace-Conditioned Skill Revision",
  "summary": "Agent skills are procedural artifacts that enable LLM agents to execute workflows, verify constraints, and recover from failures. Existing self-evolving methods refine skills using accumulated trajectories. However, they struggle in cold-start settings, where only an initial, imperfect skill is available. Consequently, skill construction defaults to expert authoring or one-shot LLM generation. Expert-authored skills are costly and may not align with how LLM agents actually execute tasks, while o",
  "authors": "Yuxuan Liu, Zhaochen Su, Lingyun Xie, Yuhao Zhang, Qing Zong, Jiahe Guo et al.",
  "category": "research",
  "topics": "agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-31T10:19:13.000Z",
  "fetched_at": "2026-07-14T16:30:09.963Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/3350",
  "original_url": "https://arxiv.org/abs/2606.01139v3",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}